US6836571B1 - Method for converting digital raster data of a first resolution into digital target data of a second resolution - Google Patents

Method for converting digital raster data of a first resolution into digital target data of a second resolution Download PDF

Info

Publication number
US6836571B1
US6836571B1 US09/555,425 US55542500A US6836571B1 US 6836571 B1 US6836571 B1 US 6836571B1 US 55542500 A US55542500 A US 55542500A US 6836571 B1 US6836571 B1 US 6836571B1
Authority
US
United States
Prior art keywords
scaling
source
smoothing
data
target
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
US09/555,425
Other languages
English (en)
Inventor
Andreas Hirn
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Canon Production Printing Germany GmbH and Co KG
Original Assignee
Oce Printing Systems GmbH and Co KG
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Oce Printing Systems GmbH and Co KG filed Critical Oce Printing Systems GmbH and Co KG
Assigned to OCE PRINTING SYSTEMS GMBH reassignment OCE PRINTING SYSTEMS GMBH ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: HIRN, ANDREAS
Application granted granted Critical
Publication of US6836571B1 publication Critical patent/US6836571B1/en
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
    • G06T3/4023—Scaling of whole images or parts thereof, e.g. expanding or contracting based on decimating pixels or lines of pixels; based on inserting pixels or lines of pixels

Definitions

  • the invention is directed to a method for converting digital source data in the raster of a first resolution into digital target data having a second resolution according to the preamble of patent claim 1 .
  • each picture element i.e. a point in the raster allocated to the digital value, is thereby referred to as pixel. Without gray levels, thus, a pixel corresponds to one bit.
  • the topical resolution is thereby indicated in picture elements per inch (dots per inch, dpi). As known, one inch corresponds to 25.4 mm.
  • the second resolution target resolution
  • the target image in the second resolution can also contain more gray scale values per pixel than the source image instead of or in addition to the higher topical resolution.
  • image data are supplied by a computer in a first raster, for example in a 240 dpi raster, but are to be reproduced by a printer in a different raster, for example in a 600 dpi raster.
  • print jobs that were produced earlier comprise, for example, only masters in 240 dpi resolution.
  • the print data must be correspondingly converted. The conversion should thereby ensue automatically without requiring inputs by the user.
  • each value of the first raster is multiplied by a scaling factor SF that is prescribed by the ratio of the two resolution values of the rasters, that, thus, for example, the SF-fold set of identical values in the second raster is generated from a value in the first raster, whereby the following applies:
  • S ⁇ ⁇ F i resolution ⁇ ⁇ of ⁇ ⁇ the ⁇ ⁇ second raster ⁇ ⁇ in ⁇ ⁇ direction ⁇ ⁇ i resolution ⁇ ⁇ of ⁇ ⁇ the ⁇ ⁇ first ⁇ raster ⁇ ⁇ in ⁇ ⁇ direction ⁇ ⁇ i ( Equation ⁇ ⁇ 1 )
  • the conversion of data into a raster having higher resolution in fact enables the improvement of the playback quality in that, for example, contours are more finely drawn. It is usually necessary to smooth the data for such a conversion.
  • a method for scaling and smoothing image data is disclosed by German Patent Document DE 195 06 792 A1.
  • a plurality of sets of pixel patterns or, respectively, Boolean calculating operations allocated to them are provided, with reference whereto the conversion ensues.
  • a matrix of source image data having, for example, 7 ⁇ 7 picture elements is subjected to the basic calculating operations and the target image data are acquired therefrom.
  • SF>1 When scaling the image data “up” (SF>1), a respective group of target pixels is allocated to a group of source pixels.
  • the calculating operations are configured such that the same number of high-resolution pixels are removed as added on average in the conversion. What is thereby achieved is that the degree of blackening of an overall image is essentially preserved.
  • a method for converting digital image data from a first raster into a second raster that is suitable for non-whole-numbered scaling factors is also disclosed by German Patent Application 197 13 079.8. This method likewise works region-oriented. A target region is thereby allocated to each source region, whereby the two regions having the same position in the overall image. Boolean calculating rules are prescribed within the target region, the conversion ensuing in conformity with these rules.
  • a scaling and smoothing of transmission data can also be necessary in the field of telefax transmission when the data, for example, are received in a first resolution but are stored, forwarded or are to be printed out in a different resolution.
  • a corresponding method for this application is disclosed, for example, by U.S. Pat. No. 5,394,485 A.
  • German Patent Document DE 42 06 277 A1 Another method for converting image data is disclosed by German Patent Document DE 42 06 277 A1. Only a raster conversion but no smoothing of the image data ensues given this method.
  • European Patent Document EP 708 415 A2 likewise discloses a method for converting image data that, however, is only suitable for whole-numbered scaling factors.
  • European Patent Document EP 0 006 351 A1 discloses an image processing system that works with look-up tables.
  • U.S. Pat. No. 5,657,430 A discloses a method for converting vector fonts onto gray scale bit maps.
  • U.S. Pat. No. 5,646,741 discloses a method and an apparatus wherein image signals are scaled and smoothed. A check according to predetermined criteria is thereby carried out in the source region to see whether a smoothing should be implemented and the source image signals potentially smoothed. The smoothed image signals are smoothed thereafter.
  • PCT Published International Application WO-A-96/16380 discloses a system and a method for the interpolation of image signals.
  • a rule is thereby respectively selected from a plurality of interpolation rules.
  • the source image signals are then processed in a plurality of successive steps.
  • the image signals are interpolated line-by-line on the basis of a selected, line-related rule.
  • the image signals are then interpolated column-by-column on the basis of a second, column-related rule.
  • the line image signals and the column image signals are compiled in pages by a formatting unit.
  • An object of the method is to provide a method for converting digital image data from a first raster into a second raster that leads to a high processing speed and that implements both a scaling as well as a smoothing of the image data.
  • This object is achieved by the invention of a method for converting digital source data referring to source pixels in the raster of a first resolution into digital target data in the raster of a second resolution, including the data are scaled by at least one scaling factor, each source datum having a target image matrix allocated to it on the basis of a surround window surrounding the source pixel and the target data being determined from neighboring target image matrices such that each target pixel is directly formed from a source pixel taking the surroundings thereof into consideration, each source datum being employed for smoothing the target data to be determined from all neighboring source data, and the scaling and the smoothing being implemented in a common processing step such that the target data are smoothed in the raster of the source data.
  • the method for converting digital source data in the raster of a first resolution into digital target data in the raster of a second resolution includes the data being scaled by a scaling factor and being smoothed, a scaling rule being prescribed from a plurality of selectable scaling rules, a smoothing rule being prescribed from a plurality of smoothing rule, a single scaling and smoothing rule being formed from the selected scaling rule and the selected smoothing rule, both a smoothing of the target data in the raster of the source data as well as a scaling ensuing in respectively one processing step with said single scaling and smoothing rule during the formation of the target data, each source datum having a target image matrix allocated to it on the basis of a surround window surrounding the source pixel and the target data being determined from neighboring target image matrices such that each target pixel is directly formed from a source pixel taking the surroundings thereof into consideration, each source datum being employed for smoothing the target data to be determined from all neighboring source data.
  • the data are scaled by at least one scaling factor and a target image matrix is allocated to each source datum by individual pixels, i.e.—pixel-individually with respect to the source pixels, on the basis of a surround window surrounding the source pixel.
  • the target data are determined from neighboring target image matrices, whereby the data are smoothed in the raster of the source data. Each source datum is thus employed for smoothing all neighboring source data.
  • the smoothing of the data is implemented in the raster of the source data and not in the target raster.
  • a significantly faster data processing given two-dimensional image data is thus possible than given comparable methods that implement the smoothing in the target raster because the data set onto which the smoothing function is applied is significantly smaller.
  • the first aspect of the invention is particularly suitable for the conversion of image data given a non-whole-numbered broken) scaling factor.
  • the first aspect of the invention is based on the perception that the same result can be achieved with a smoothing in the source raster as with a smoothing that is applied to a significantly greater plurality of data in the target raster because the structures to be smoothed are already to be defined from the source image.
  • the scaling of an image by a factor greater than one increases the plurality of pixels to be smoothed; the informational content of the bit map on which the image is based, however, remains unmodified. Tests have shown that a smoothing with rules that erected in the target raster does not yield different results than when corresponding rules for smoothing are already erected on the basis of the data in the source raster.
  • a smoothing with the data of the source image as basis also enables a smaller size of the recognition matrix.
  • a recognition matrix of 3 ⁇ 3 in the source region for example, achieves the same quality as a 5 ⁇ 5 recognition matrix that is applied in the target region.
  • the processing speed of the invention method given direct logical evaluation is thus increased in two respects: first, fewer data are to be interpreted in the source raster than in the target raster; second, the size of the smoothing window can be reduced in the source raster.
  • the processing speed is then higher by a factor of up to 25/9 ⁇ SF x ⁇ SF y than in conventional methods.
  • the logical outlay for example for gate functions, is reduced by this factor.
  • a table having 512 entries is needed for a 3 ⁇ 3 matrix given a realization with look-up tables, which are often utilized in software solutions for performance-enhancement because the bit-by-bit logical interpretation is thereby eliminated and the result is directly obtained from the table.
  • this table must have a size of 33554432 entries (32 MB) given a 5 ⁇ 5 matrix. A table of this size is no longer acceptable in practice.
  • the invention also enables both the function of the smoothing as well as that of the scaling to be implemented in a single step, in that the overall method is implemented in the raster of the source data.
  • the method can thereby be implemented independently of the size of the respective scaling factor.
  • the scaling factor can be both whole-numbered as well as fractional.
  • digital source data in the raster of a first resolution are scaled by a scaling factor and smoothed into digital target data in the raster of a second resolution.
  • a scaling rule is thereby prescribed and a specific smoothing rule is prescribed from a plurality of smoothing rules.
  • the two prescribed rules are then merged such to form a combined scaling and smoothing rule that the smoothing ensues in the raster of the source data, whereby each source datum is employed for smoothing a plurality of neighboring source data.
  • the scaling factor is, in particular, not a whole number and can be presented by a fraction of whole numbers.
  • a high degree of flexibility in the processing of image data is achieved by the second aspect of the invention.
  • a plurality of smoothing and/or scaling methods can thereby be freely combined with one another, and one can react very flexibly to the greatest variety of print data and printer resolutions when printing images.
  • Individual (job-specific) scaling and/or smoothing rules can thereby already be prescribed or selected either in the print job or in the printer device, for example by an operator.
  • a third aspect of the invention it is not only binary data (black-and-white) that are processed; rather, grayscale values or color values covering a plurality of bits or bytes are processed per picture element. It is thereby possible, on the one hand, to implement a “grayscale conversion” wherein the raster refers to gray scales and, thus, a conversion from a first grayscale raster into a second grayscale raster is undertaken per picture element, for example 4 bit grayscale values corresponding to 16 gray scales onto 6-bit grayscale values corresponding to 64 gray scales are scaled up. A grayscale smoothing can thereby also ensue in that more finely graduated grayscale transitions are generated between the picture elements in the target space.
  • the scalings and smoothings in the location space, in the grayscale space and in the color space can thereby be arbitrarily combined with one another.
  • the processing of the data ensues byte-oriented.
  • a binary information can thereby respectively be allocated to a plurality of picture elements and the data can be processed parallel.
  • gray scales and/or color values can also be allocated to the picture elements (pixels), these in turn comprising a plurality of bits or bytes per pixel.
  • a byte-by-byte processing has a positive effect on the processing speed because digital electronic components, particularly in the field of information processing, likewise internally process the data byte-by-byte and because this byte format is a generally standard memory format.
  • the data are thereby shifted in a register by a specific plurality of positions dependent on the height of the smoothing window; after storing a corresponding plurality of bytes (for example, 3 bytes for a processing of 3 lines with 8 pixels each on which a 3 ⁇ 3 smoothing window should respectively act), neighboring data represent an index.
  • This index can be directly employed for addressing a corresponding smoothing matrix (for example, 3 ⁇ 3), whereby the addressing acts either as input signal of a hardware circuit or acts directly on a look-up table within a computer software.
  • the two-dimensional objective of processing image data is thereby converted into a one-dimensional task.
  • a shift register having respectively n bytes per line is filled per processing clock according to the following rules:
  • R (i+A) q(i/Q y , Q y ⁇ 1 ⁇ (i%Q y )) or
  • R i value of the i th register pixel
  • W value of a pixel, i.e. bits per pixel (binary, grayscale value, color value)
  • the inventive method runs significantly faster than comparable methods that first implement a scaling, deposit the result in an intermediate memory and only then implement the smoothing at the intermediately stored data, i.e. in the target raster.
  • What is advantageous given a conversion with software is that the switching can ensue highly suited to need within a print job —when a conversion is required, this ensues with the corresponding modules of the conversion program. When no conversion is required, then the data are forwarded without having been processed by the conversion program.
  • the flexibility can thus be enhanced to such an extent that different resolutions can even be processed within one document to be printed out, i.e. within one page. Whereas, for example, text having a resolution of 300 dpi has a good effect, it is usually expedient to select a resolution of 600 dpi or higher in the reproduction of images.
  • the scaling and smoothing can ensue in a common step with a look-up table that contains data for both procedures.
  • the source data are thereby preferably directly employed for addressing the look-up table.
  • FIG. 1 a procedure of the prior art
  • FIG. 2 a mathematical model on which the invention is based
  • FIG. 3 various combination possibilities in an image data conversion of 2 ⁇ 2 source pixels
  • FIG. 4 an example of an image data conversion
  • FIG. 5 a further example of an image data conversion
  • FIG. 6 various presentations of a slanting line in an image raster
  • FIG. 7 a smoothing window in an image raster
  • FIG. 8 a scaling procedure by a scaling factor of 2
  • FIG. 9 a smoothing procedure with a 5 ⁇ 5 matrix
  • FIG. 10 a scaling by the factor 2.5
  • FIG. 11 a sketch underlying a smoothing procedure
  • FIG. 12 various windows on which a smoothing is based
  • FIG. 13 the schematic diagram of a smoothing result
  • FIG. 14 an illustration of the superimposition of data
  • FIG. 15 the processing of image data with a smoothing window
  • FIG. 16 the deposit of two-dimensional image data in a one-dimensional register
  • FIG. 17 the conversion of a plurality of pixels of a source line into register pixels
  • FIG. 18 a data processing process, whereby scaled and smoothed target image data are acquired directly from the source image data
  • FIG. 19 a conversion of digital image data into index bits
  • FIG. 20 a hardware arrangement for the conversion of digital image data
  • FIG. 21 a software concept for the conversion of digital image data
  • FIG. 22 a version for compiling target image matrices without superimposition of source pixels
  • FIG. 23 the result of the compilation of FIG. 22
  • FIG. 2 b shows a scaling by a scaling factor 2 as occurs given a conversion from a 300 dpi source raster onto a 600 dpi target raster.
  • Each source pixel 7 is thereby two-dimensionally treated, i.e. doubled in each of the directions x and y.
  • the raster spacings are twice as great in the source raster as in the target raster.
  • One pixel 7 in the source region becomes four pixels in the target region.
  • the scaling factor can also be different in the x-and y-directions, for example have the value 2 in x-direction and the value 3 in y-direction.
  • a scaling factor that is not a whole number is to form the basis, for example the scaling factor 2.5 corresponding to a conversion from a 240 dpi source raster onto a 600 dpi target raster, then one proceeds analogous to whole-numbered scaling factors. This procedure is schematically shown in FIG. 2 d . Theoretically, 2.5 ⁇ 2.5 pixels 8 in the target region derive from one pixel 7 in the source region.
  • this condition is met, for example, with two source pixels and 5 target pixels.
  • this condition is met, for example, with two source pixels and 5 target pixels.
  • one proceeds on the basis of a 2 ⁇ 2 pixel square in the source region, then one obtains a 5 ⁇ 5 pixel square in the target region given the scaling factor 2.5.
  • FIGS. 10 through 13 An example for a conversion having a scaling factor that is not a whole number is indicated below (FIGS. 10 through 13 ).
  • a source pixel is imaged onto a rectangle having sx*sy target pixels, i.e. a plurality of target pixels are derived from one source pixel.
  • the target pixels contain the same value (9/1 given binary data, gray scales or, respectively, color value given non-binary data) as the source pixel.
  • each target pixel is derived from one source pixel.
  • the plurality of target pixels that are derived from a specific source pixel is thereby dependent on the location of the target pixel and is not always the same; an asymmetry therefore arises.
  • An improvement compared to the first scaling method is obtained when a rectangle of sx N *sy N source pixels are combined to a block 7 ′ that can be represented in the target region without sub-pixels.
  • Such rectangles are scaled block-by-block into corresponding target blocks 8 ′ with sx Z ⁇ sy Z , in that each target pixel is derived from the source pixels from the source block via a logical equation.
  • Each target pixel can then be dependent on a plurality or, respectively, on from one through all source pixels. The conversion described for FIG. 4 could then look the way it is shown in FIG. 5 .
  • each pixel is observed in its environment.
  • a quadratic smoothing window smoothing matrix
  • Dependent on the environment and its own pixel value (the recognized structure in the smoothing window) a decision is made as to whether the pixel should be black or white given binary data; the resulting value is determined given grayscale values or color values.
  • Only pixel values are usually shifted in sum over all pixels; the sum of the pixels that are set or, respectively, not set with the digital values “0” or, respectively, “1” or, respectively, the average grayscale value or color value remains nearly the same.
  • the smoothing ensues by attaching and removing pixels; the structures to be recognized and to be corrected are also referred to as rules.
  • the size of the environment to be considered is dependent on the prescription as to which structures are to be recognized and smoothed.
  • Neighbors of the first order are the direct neighbors, i.e. all pixels that share at least one corner with the pixel to be investigated. Overall, this yields the investigated pixel and eight neighbors.
  • Neighbors of the second order are all pixels that share at least one corner with the neighbors of the first order, etc.
  • FIG. 6 c The same result (FIG. 6 c ) can be achieved when, given the image excerpt shown in FIG. 6 a , corner pixels at black-to-white transitions are removed from left to right. This latter method is single-stage because adding and removing are implemented in one work step.
  • sm ( i g , j g ) ⁇ fsmooth ( q ( i 2 - 1 , j 2 - 1 ) , q ( i 2 - 1 , j 2 - 1 ) , q ( i 2 , j 2 - 1 ) , q ( i 2 , j 2 - 1 ) ⁇ q ( i 2 + 1 , j 2 - 1 ) , ⁇ q ( i 2 - 1 , j 2 - 1 ) , q ( i 2 - 1 , j 2 - 1 ) , q ( i 2 , j 2 - 1 ) , q ( i 2 , j 2 - 1 ) , q ( i 2 , j 2 - 1 ) , q ( i 2 , j 2 - 1 ) ⁇ q ( i 2 + 1 , j 2 - 1
  • sm ( i u , j g ) ⁇ fsmooth ( q ( i 2 - 1 , j 2 - 1 ) , q ( i 2 , j 2 - 1 ) , q ( i 2 , j 2 - 1 ) , q ( i 2 + 1 , j 2 - 1 ) , q ( i 2 + 1 , j 2 - 1 ) , ⁇ q ( i 2 - 1 , j 2 - 1 ) , q ( i 2 , j 2 - 1 ) , q ( i 2 , j 2 - 1 ) , q ( i 2 + 1 , j 2 - 1 ) , q ( i 2 + 1 , j 2 - 1 ) , q ( i 2 + 1 , j 2 - 1 ) , q ( i 2 + 1 , j 2
  • four target pixels can be determined from nine source elements or, respectively, nine source pixels.
  • the four target pixels are in fact calculated from the same source pixels
  • a matrix having the size 3 suffices in the source region, i.e. the equations in the source region contain only 9 dependents instead of 25 in the target region.
  • the calculating outlay is in fact increased but the speed is also increased; the four target pixels are thereby calculated parallel and independently of one another.
  • a combination of specific, logical sub-operations is possible dependent on the equations.
  • the scaling is contained in the process; the data need be processed only once not twice as in the above-described, known methods.
  • the smoothing scaling in the source raster enables a faster implementation of the operations given the same quality as in the two-stage process.
  • the relative gain as a result of the single-stage method for smoothing and scaling is all the greater the greater the scaling factor is.
  • the rectangle is drawn with the target pixels (target rectangle).
  • the smoothing matrix having the edge length G in the target region is placed around each corner target pixel.
  • the size of the source rectangle to be considered can be determined from the expanse.
  • All target pixels in the target rectangle can be calculated from the source pixels in the source rectangle. Only the source matrix need be covered for the calculation.
  • the pixels in the target rectangle can be determined parallel independently from one another.
  • a smoothing with a filter window 17 having the size 5 ⁇ 5 is now to be implemented within the region 16 .
  • This 5 ⁇ 5 window can be accommodated four times in the region 16 .
  • FIGS. 9 a and 9 c and 9 d Each of the operations shown in FIGS. 9 a through 9 d can be computationally represented by a recognition matrix, whereby respectively one pixel value is determined per matrix. For example, the value of the pixel 18 is calculated from the position of the filter window 17 shown in FIG. 9 a.
  • FIGS. 9 a through 9 d Four smoothed target pixels 18 , 19 , 20 and 21 are obtained from the recognition matrices on which FIGS. 9 a through 9 d are based. This group of target pixels is referenced in common as 22 in FIG. 9 e.
  • the 3 ⁇ 3 source matrix 15 describes 512 possible pixel combinations.
  • An intermediate matrix from which four target pixels then derive with a 5 ⁇ 5 smoothing can be determined for each of these combinations with a general scaling method (scaling factor 2).
  • scaling and the smoothing can be implemented in one step, since an unambiguous relationship between source matrix (central pixel with 8 surrounding pixels) and the target pixels (target matrix) exists.
  • Each source pixel is thereby directly converted into four target pixels taking its surroundings into consideration.
  • a table is employed for each step, 9 target pixels being produced from 9 source pixels therewith. To that end, 4 ⁇ 512 entries of 9 bits each or, respectively, a table size of 4096 bytes is required. Compared to the aforementioned 262144 bytes, this is a memory reduction by the factor 64. The 3 ⁇ 3 target pixels generated in this way are then placed on top of one another in the form of an OR-operation.
  • FIGS. 12 and 13 illustrate this procedure: in FIG. 12, four source pixels 2 — 2 , 2 - 3 , 3 - 2 or, respectively, 3 — 3 are shown with their respective surrounding source pixels, i.e. source pixel windows 52 a , 52 b , 52 c , 52 d .
  • the source pixels are present in a 240 dpi raster.
  • a respective 3 ⁇ 3 target pixel square (matrix) 26 , 27 , 28 or, respectively, 29 are to be respectively formed from the source pixel 2 — 2 , 2 - 3 , 3 - 2 and 3 — 3 , for example the target pixel square 26 for source pixel 2 — 2 , the target pixel square 27 for source pixel 2 - 3 , etc.
  • the target pixel squares 26 , 27 , 28 , 29 are then placed on top of one another such that the lines 31 , 31 ′, 31 ′′ and 31 ′′′ respectively align with one another, as do the lines 32 , 32 ′, 32 ′′ and 32 ′′′. Due to this overlay, the same source pixels ( 1 - 2 , 1 - 3 ; 2 - 1 . . . 2 - 4 ; 3 - 1 . . . 3 - 4 ; 4 - 2 , 4 - 3 ) of the source pixel windows 52 a . . . 52 d lie congruently on top of one another. Further, the 5 ⁇ 5 target pixel square 30 of FIG.
  • the first and all following odd-numbered source pixels are converted according to the source pixel window 52 a (forming a respective target pixel matrix of the type 26 ); the second and all following even-numbered source pixels of these lines are converted according to source pixel window 52 b (forming a respective target pixel matrix of the type 27 ).
  • the first and all odd-numbered successor source pixels are respectively converted according to the source pixel window 52 c into a target pixel matrix of the type 28 , and the even-numbered source pixels are converted according to the source pixel window 52 d into a target pixel matrix of the type 29 .
  • asymmetrical definition and joining by insertion into one another (overlapping) of the source pixels an asymmetrical definition and joining is also possible according to FIGS. 22 and 23.
  • a 3 ⁇ 3 target pixel square 53 is thereby formed from the source pixel 2 — 2
  • a 2 ⁇ 3 target pixel square 54 is formed from the source pixel 2 - 3
  • a 3 ⁇ 2 target pixel square 55 is formed from the source pixel 3 - 2
  • a 2 ⁇ 2 target pixel square 56 is formed from the source pixel 3 — 3 .
  • the target pixel rectangles or, respectively, squares 53 , 54 , 55 and 56 (target image matrices) formed as a result thereof are then joined as target image without overlap.
  • the other method steps cited in the symmetrical processing (for example, register entry, index formation, line-by-line procedure) are thereby identically implemented.
  • Source pixels are generally present as source image in a memory area that is organized in rows byte-oriented. A stored row thereby corresponds to an image line (scan line).
  • FIGS. 15 through 18 A byte-oriented procedure with which the source pixels are one-dimensionally processed with the assistance of a shift register is illustrated on the basis of FIGS. 15 through 18.
  • the source pixels are entered into the shift register according to specific conventions, as shown in FIG. 16 . This can happen directly in hardware in that the individual lines are transferred to the corresponding locations in the register. Given a realization in software, a look-up table is employed for performance reasons. This table is constructed according to FIG. 17 .
  • the upper scan line is directly copied into the register after conversion, the next scan line is shifted one pixel position toward the left after conversion and is then entered into the register, the third scan line is entered into the register shifted by two positions toward the left after conversion (FIG. 18 ).
  • FIG. 15 a shows an excerpt 33 , the three stripes 35 , 35 ′ and 35 ′′ of the source image that are respectively 8 pixels wide and lie above one another and are limited by the lines 34 and 34 ′.
  • the respectively last pixel of the proceeding 8 pixels are shown at the left next to this and the respectively first pixel of the next 8 pixels are also shown to the right thereof
  • a 3 ⁇ 3 recognition window 36 is shifted across this structure from left to right.
  • the sequence of the FIGS. 15 a , 15 b , 15 c and 15 d illustrates this for the first four shift events.
  • This index can be directly taken from the register 37 as the value formed from the region 38 (the nine right-hand, neighboring bits of the register 37 ).
  • the bits of this value correspond to the recognition window and yield the smoothed target pixels.
  • the index data can be constructed once for each block of 3 bytes source data in a register.
  • the determination of the indices for the scaling/smoothing matrices of an image line can then be optimized even more.
  • the structure of the index register from the respective byte, namely, can then be realized in only one step via a table that has the properties schematically shown in FIG. 17 .
  • the lines 39 respectively connect the left edge of a source 40 to the register position 41 appertaining thereto.
  • the conversion table is then simply shifted one pixel (the middle pixel) or two pixels (the lower byte) toward the left and is ordered onto the index register.
  • This conversion table is referred to below as index table.
  • FIG. 18 again illustrates the entire procedure for smoothing and scaling the source image data based on a byte-by-byte processing and programmed in software.
  • the 30-bit register 37 is filled (whereby, of course, a 32 bit register can also be employed) via the look-up table from respectively three bytes input image data 42 in the excerpt 33 of the source image.
  • a respective shift by one position is undertaken in the processing step 44 ( ⁇ 1, ⁇ 2).
  • Register locations that have already been written with the preceding byte are overwritten with the following data with an OR-operation.
  • the lower 9 bytes 45 of the register 37 yield an index for the scaling/smoothing table 46 from which the scaled and smoothed target pixels 47 can be directly taken. These are then deposited in the target region. Subsequently, the next 3-byte block in the source region is processed. This procedure is repeated moving across the entire source image. Pixels that have not been set are assumed at the edges for the edge pixels that are not present.
  • the shift register is filled according to the following equation in each step with respectively n bytes of each scan line (line) in the source pixel rectangle:
  • n plurality of bytes read in per line
  • W value of a pixel, i.e. bits per pixel (binary, gray scale value, color value), and
  • FIG. 20 shows a corresponding realization in hardware.
  • the respectively obtained index bits 49 form the input signals for a logic circuit 50 for calculating the target pixels.
  • FIG. 21 schematically shows a conversion in the form of software.
  • the index bits 49 here serve as index for addressing a look-up table 51 that contains the previously calculated pixels for this combination.
  • the processing of grayscale of color pixels ensues according to the same principle as with binary data. All boxes in the illustrated figures then represent one pixel that contains W bits per pixel instead of one bit per pixel given binary data.
  • the index formation as well as the equations for the combination of scaling and smoothing all refer to pixels; the only thing changed is the plurality of bits per pixel.
  • the scaling/smoothing table then contains grayscale or, respectively, color values instead of bits per pixel.
  • the shift values onto the index register ( ⁇ 1, ⁇ 2) indicate pixel positions; expressed in bits, this is ( ⁇ W, ⁇ 2 ⁇ W) bits.
  • the method can thus also be used for the processing of data having W bits per pixel in order to convert data of a first raster into a second that is finer (scaling up).
  • Scaling factor 2 we obtain 4 target pixels for a source pixel.
  • the four acquired pixels have values between b0000 and b1111, whereby the preceding b indicates binary notation.
  • These pixels are now converted into gray scales, whereby the plurality of black pixels (for example, having pixel value 1) is summed up and converted into a grayscale value.
  • the conversion need not ensue linearly, it can also be based on the (non-) linearity of the output unit.
  • the conversion ensues with a table, for example
  • Pixel Value Grayscale Value No black pixel b0000 b00 one black pixel b0001 b01 b0010 b01 b0100 b01 b1000 b01 two black pixels b0011 b10 b0101 b10 b1001 b10 b0110 b10 b1010 b10 b1100 b10 three black pixels b0111 b11 b1011 b11 b1101 b11 b1110 b11 four black pixels b1111 b11
  • the invention was specifically described for employment in a printer that converts the image data from a first raster into a second raster upon retention or enhancement of the gray scales or, respectively, color values. Only an enhancement of gray scales/color steps in the same raster is also possible. It is thereby clear that the image data can also be edited such within a computer that they are available in a resolution adapted to the printer. Particularly in a network wherein print jobs from various computers are sent to a central printer, this will generally be the case. The conversion can thereby ensue both in the sending computer as well as in an intervening computer that administers the print jobs.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Editing Of Facsimile Originals (AREA)
  • Image Processing (AREA)
  • Facsimile Image Signal Circuits (AREA)
US09/555,425 1997-11-28 1998-11-27 Method for converting digital raster data of a first resolution into digital target data of a second resolution Expired - Lifetime US6836571B1 (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
DE19752927 1997-11-28
DE19752927 1997-11-28
PCT/EP1998/007689 WO1999028864A1 (de) 1997-11-28 1998-11-27 Verfahren zur umsetzung digitaler daten im raster einer ersten auflösung in digitale zieldaten einer zweiten auflösung

Publications (1)

Publication Number Publication Date
US6836571B1 true US6836571B1 (en) 2004-12-28

Family

ID=7850188

Family Applications (1)

Application Number Title Priority Date Filing Date
US09/555,425 Expired - Lifetime US6836571B1 (en) 1997-11-28 1998-11-27 Method for converting digital raster data of a first resolution into digital target data of a second resolution

Country Status (5)

Country Link
US (1) US6836571B1 (ja)
EP (1) EP1034511B1 (ja)
JP (1) JP4156194B2 (ja)
DE (1) DE59806214D1 (ja)
WO (1) WO1999028864A1 (ja)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020009237A1 (en) * 2000-07-21 2002-01-24 Tadanori Tezuka Display reduction method using sub-pixels
US20020167678A1 (en) * 2001-05-11 2002-11-14 Fujitsu Limited Image processing method, image processing equipment and image forming equipment
US20040057634A1 (en) * 2002-07-10 2004-03-25 Gennosuke Mutoh Image processing apparatus, image processing method and program
US20060215186A1 (en) * 2005-03-22 2006-09-28 International Business Machines Corporation Method and system for scaling with dot gain control
WO2008030163A1 (en) * 2006-09-08 2008-03-13 Telefonaktiebolaget Lm Ericsson (Publ) Image scaling method

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102063123B1 (ko) * 2018-05-23 2020-01-07 서울대학교산학협력단 의존성 그래프를 이용한 데이터 처리 시스템 및 방법

Citations (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4029947A (en) * 1973-05-11 1977-06-14 Rockwell International Corporation Character generating method and system
EP0006351A1 (en) 1978-06-21 1980-01-09 Xerox Corporation Method of processing image data
DE4027897A1 (de) 1989-09-05 1991-03-21 Canon Kk Bildverarbeitungsvorrichtung
DE4206277A1 (de) 1991-12-07 1993-06-09 Samsung Electronics Co., Ltd., Suwon, Kr Aufloesungs-wandel-verfahren fuer ein bildverarbeitungssystem
US5270836A (en) 1992-11-25 1993-12-14 Xerox Corporation Resolution conversion of bitmap images
US5394485A (en) 1993-03-17 1995-02-28 Hewlett-Packard Company Method and apparatus for smoothed scaling of facsimile images
US5440407A (en) * 1994-03-11 1995-08-08 Hewlett-Packard Company Pixel correction and smoothing method
EP0708415A2 (en) 1994-10-18 1996-04-24 Hewlett-Packard Company Four-quadrant scaling of dot matrix data
WO1996016380A1 (en) 1994-11-23 1996-05-30 Minnesota Mining And Manufacturing Company System and method for adaptive interpolation of image data
DE19506792A1 (de) 1995-02-27 1996-09-05 Siemens Nixdorf Inf Syst Verfahren und Vorrichtung zur Steigerung der Bildqualität in Bildausgabegeräten
EP0506379B1 (en) 1991-03-28 1997-05-21 Canon Kabushiki Kaisha Image processing apparatus and method
US5646741A (en) 1994-02-25 1997-07-08 Canon Kabushiki Kaisha Image processing apparatus and method capable of smoothing image data in accordance with an image enlargement/reduction ratio
US5655061A (en) * 1995-11-15 1997-08-05 Xerox Corporation System and method for providing a high addressable printing system
US5793379A (en) * 1995-04-03 1998-08-11 Nvidia Corporation Method and apparatus for scaling images having a plurality of scan lines of pixel data
US5872572A (en) * 1995-12-08 1999-02-16 International Business Machines Corporation Method and apparatus for generating non-uniform resolution image data
US5982508A (en) * 1995-05-23 1999-11-09 Canon Kabushiki Kaisha Image processing method and apparatus
US6002446A (en) * 1997-02-24 1999-12-14 Paradise Electronics, Inc. Method and apparatus for upscaling an image
US6034786A (en) * 1996-09-02 2000-03-07 Samsung Electronics Co., Ltd. Apparatus and method for enlarging or reducing an image in an image processing system
US6091859A (en) * 1993-02-12 2000-07-18 Fuji Xerox Co., Ltd. Image data enlarging/smoothing processor

Patent Citations (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4029947A (en) * 1973-05-11 1977-06-14 Rockwell International Corporation Character generating method and system
EP0006351A1 (en) 1978-06-21 1980-01-09 Xerox Corporation Method of processing image data
DE4027897A1 (de) 1989-09-05 1991-03-21 Canon Kk Bildverarbeitungsvorrichtung
EP0506379B1 (en) 1991-03-28 1997-05-21 Canon Kabushiki Kaisha Image processing apparatus and method
DE4206277A1 (de) 1991-12-07 1993-06-09 Samsung Electronics Co., Ltd., Suwon, Kr Aufloesungs-wandel-verfahren fuer ein bildverarbeitungssystem
US5270836A (en) 1992-11-25 1993-12-14 Xerox Corporation Resolution conversion of bitmap images
US6091859A (en) * 1993-02-12 2000-07-18 Fuji Xerox Co., Ltd. Image data enlarging/smoothing processor
US5394485A (en) 1993-03-17 1995-02-28 Hewlett-Packard Company Method and apparatus for smoothed scaling of facsimile images
US5646741A (en) 1994-02-25 1997-07-08 Canon Kabushiki Kaisha Image processing apparatus and method capable of smoothing image data in accordance with an image enlargement/reduction ratio
US5440407A (en) * 1994-03-11 1995-08-08 Hewlett-Packard Company Pixel correction and smoothing method
EP0708415A2 (en) 1994-10-18 1996-04-24 Hewlett-Packard Company Four-quadrant scaling of dot matrix data
US5757982A (en) * 1994-10-18 1998-05-26 Hewlett-Packard Company Quadrantal scaling of dot matrix data
WO1996016380A1 (en) 1994-11-23 1996-05-30 Minnesota Mining And Manufacturing Company System and method for adaptive interpolation of image data
DE19506792A1 (de) 1995-02-27 1996-09-05 Siemens Nixdorf Inf Syst Verfahren und Vorrichtung zur Steigerung der Bildqualität in Bildausgabegeräten
US5793379A (en) * 1995-04-03 1998-08-11 Nvidia Corporation Method and apparatus for scaling images having a plurality of scan lines of pixel data
US5982508A (en) * 1995-05-23 1999-11-09 Canon Kabushiki Kaisha Image processing method and apparatus
US5655061A (en) * 1995-11-15 1997-08-05 Xerox Corporation System and method for providing a high addressable printing system
US5872572A (en) * 1995-12-08 1999-02-16 International Business Machines Corporation Method and apparatus for generating non-uniform resolution image data
US6034786A (en) * 1996-09-02 2000-03-07 Samsung Electronics Co., Ltd. Apparatus and method for enlarging or reducing an image in an image processing system
US6002446A (en) * 1997-02-24 1999-12-14 Paradise Electronics, Inc. Method and apparatus for upscaling an image

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
A. Iwainsky et al., LEXIKON Der Computergrafik und Bildverarbeitung.
IBM Technical Disclosure Bulletin, "Fast Scaling Method to Reduce Binary Images by Specific Factors", pp. 357-359.

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020009237A1 (en) * 2000-07-21 2002-01-24 Tadanori Tezuka Display reduction method using sub-pixels
US20020167678A1 (en) * 2001-05-11 2002-11-14 Fujitsu Limited Image processing method, image processing equipment and image forming equipment
US20040057634A1 (en) * 2002-07-10 2004-03-25 Gennosuke Mutoh Image processing apparatus, image processing method and program
US7414639B2 (en) 2002-07-10 2008-08-19 Ricoh Company, Ltd. Image processing apparatus, image processing method and program
US20060215186A1 (en) * 2005-03-22 2006-09-28 International Business Machines Corporation Method and system for scaling with dot gain control
US8520253B2 (en) * 2005-03-22 2013-08-27 Infoprint Solutions Company, Llc Method and system for scaling with dot gain control
WO2008030163A1 (en) * 2006-09-08 2008-03-13 Telefonaktiebolaget Lm Ericsson (Publ) Image scaling method
CN101512595B (zh) * 2006-09-08 2012-06-20 艾利森电话股份有限公司 图像缩放方法

Also Published As

Publication number Publication date
WO1999028864A1 (de) 1999-06-10
JP4156194B2 (ja) 2008-09-24
EP1034511B1 (de) 2002-11-06
DE59806214D1 (de) 2002-12-12
JP2001525623A (ja) 2001-12-11
EP1034511A1 (de) 2000-09-13

Similar Documents

Publication Publication Date Title
US6901170B1 (en) Image processing device and recording medium
DE69611573T2 (de) Auflösungserhöhung mit Standardvergleichsmustern
US5282057A (en) Bit-map image resolution converter
EP0115584B1 (en) Image producing apparatus and methods of processing image-representing signals for use by such apparatus
US8331731B2 (en) Image processing method and image processing apparatus
EP1429286A1 (en) Method of preparing bit map
DE69117100T2 (de) Verfahren und Gerät zur Erzeugung von ein integriertes Bild darstellenden Bilddaten
JPH03214271A (ja) テキストの画素イメージの質を改善する方法
EP0407614B1 (en) Image processing apparatus
US5583953A (en) Intelligent doubling for low-cost image buffers
KR100438246B1 (ko) 스케일링시스템
US6836571B1 (en) Method for converting digital raster data of a first resolution into digital target data of a second resolution
JP3990045B2 (ja) 情報処理方法、情報処理装置及び記録媒体
JPH03105687A (ja) 画像合成方法
DE69624745T2 (de) Verfahren und System zur Umsetzung von Multigradationsbilddaten in Binärbilddaten
US5526474A (en) Image drawing with improved process for area ratio of pixel
US5394485A (en) Method and apparatus for smoothed scaling of facsimile images
JPH0253761B2 (ja)
US6567565B1 (en) Antialiased image rendering algorithm
JP2002507848A (ja) ランダムにシフトされるピクセルを有する、画像データの座標変換方法
JP2000351242A (ja) 画像形成装置
JP3115701B2 (ja) 画像処理装置
JP2950684B2 (ja) 画像縮小装置
EP0439714A2 (en) Anti-aliasing method for graphic display
JP3005014B2 (ja) 印刷装置

Legal Events

Date Code Title Description
AS Assignment

Owner name: OCE PRINTING SYSTEMS GMBH, GERMANY

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:HIRN, ANDREAS;REEL/FRAME:011935/0044

Effective date: 20000517

STCF Information on status: patent grant

Free format text: PATENTED CASE

FPAY Fee payment

Year of fee payment: 4

FPAY Fee payment

Year of fee payment: 8

FPAY Fee payment

Year of fee payment: 12